MIP-Guided LFNST Scanning for Flexible VVC Intra Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The lack of variability in scanning order selection for Low-Frequency Non-Separable Transform (LFNST) technology in non-traditional intra prediction modes reduces encoding efficiency in video coding standards like H.266/VVC.
Innovation Solution
Introduce a Matrix-based Intra Prediction (MIP) parameter to determine a flexible scanning order for LFNST coefficients, allowing for horizontal and vertical scanning orders, enhancing the applicability of LFNST to non-traditional intra prediction modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If LFNST technology is applied to non-traditional intra prediction modes with fixed scanning order selection, then the technology can be applied to these modes, but the encoding efficiency is reduced due to lack of variability in scanning order selection
Solution Approach 1:
The patent introduces dynamic scanning order selection for LFNST in non-traditional intra prediction modes by utilizing MIP parameters. Instead of using a fixed scanning order, the method dynamically determines the scanning order based on the MIP mode index and block dimensions, allowing the system to adapt to different prediction modes and block characteristics, thereby improving encoding efficiency while maintaining broad applicability.
Solution Approach 2:
The patent changes the parameter used for scanning order selection from traditional intra prediction mode indices to MIP-specific parameters (mipModeIndex and block dimensions). This parameter change enables the LFNST technology to work effectively with non-traditional intra prediction modes by using parameters that are meaningful and variable within the MIP framework, thus resolving the contradiction between adaptability and encoding efficiency.
2Ease of manufacture
If traditional intra prediction mode-based scanning order selection is used, then the method is simple to implement, but it lacks variability for non-traditional modes like MIP
Solution Approach 1:
The patent creates a universal scanning order selection mechanism that works for both traditional and non-traditional intra prediction modes. By using MIP parameters (mipModeIndex, block width and height) that are available across different mode types, the method achieves multi-functionality where the same LFNST process can adapt to various prediction modes including MIP, planar, and angular modes, thereby increasing variability without sacrificing implementation simplicity.
Data Source
AI summary
A transform method includes: determining a prediction mode parameter of a current block; determining a MIP parameter when the prediction mode parameter indicates that MIP is used for the current block to determine an intra prediction value; determining the intra prediction value of the current block according to the MIP parameter, and calculating a residual value between the current block and the intra prediction value; performing a first transform on the residual value to obtain a first coefficient matrix; determining a scanning order of LFNST coefficients used for the current block according to the MIP parameter when an LFNST is used for the current block; constructing an input coefficient matrix of the LFNST based on the first coefficient matrix according to the scanning order of LFNST coefficients; and performing an LFNST processing on the input coefficient matrix to obtain a transform coefficient matrix of the current block.